What is knowledge-wiki?

knowledge-wiki is a tool that pulls a team's scattered documentation, spreadsheets, API specs, code comments, and meeting notes into one structured, searchable knowledge base that an AI agent can search and reason over directly. The problem it targets is familiar to many teams: when someone asks how a piece of the system works or why a past decision was made, people end up digging through several different tools or asking a colleague who may not remember either. It works as a set of commands, prefixed with a slash, that you run through an AI coding assistant. An initialize command sets up a folder structure, generates governance files like CLAUDE.md and CONTRIBUTING.md, installs supporting scripts, and adds a pre push git hook that checks for outdated documentation before you push code. From there you import documents from sources including Lark, Apipost, local code folders, and PDF files, and register external sources so the tool can automatically detect when they change and flag documentation that may be out of date. Once your knowledge base is built, you can ask it questions directly and get answers pulled from your own content, or ask it to reason step by step through more complex, multi part questions. It combines three retrieval techniques, keyword based search, dense vector search, and graph based search, to find relevant information. It can also auto generate wiki style pages that link to each other based on the imported content. The project supports over eleven input formats and documents both an English and a Chinese version of its README. It is aimed at engineering teams who want their internal knowledge to stay current and to be answerable by AI agents rather than scattered and forgotten across many separate tools.